After some trial and error, I came across http://r.research.att.com/tools/#gcc42 that describes a process of building gfortran using Xcode gcc-4.2 (available since Xcode release 3.1.1, at least --- current is Xcode 3.1.4 released Sept 2009). This will work on both PPC and Intel Macs running at MacOSX version 10.5 or higher. I imagine this can be put into the fortran spkg. There are also instructions for MacOSX 10.4 (the Apple's gcc is older there, so the patch is different...)
So yes, it seems that g95 can be dropped this way, but this is a considerable amount of work. The amount of work would decrease if we drop MacOSX 10.4 from the list of supported OSes... Instead of this, it looks easier to require gfortran from the above be installed (they provide binaries), just as gfortran is required on Linux. I am trying out the latter route right now, and will report on it here. Dima On Sep 10, 1:03 am, kcrisman <kcris...@gmail.com> wrote: > On Sep 9, 12:31 pm, Dima Pasechnik <dimp...@gmail.com> wrote: > > > I didn't build sage on my G4 since 4.3.4, and there gfortran points to > > something I got from fink. > > So I must have built Sage using fink's gfortran, built against > > gcc-4.3.4 ? Hmm. > > OK, let me try the current release and see if I get anywhere. > > (It will take a while...) > > That should be ok, though. The point is that you used something in > fink, and you didn't use Xcode's GCC. This is a useful data point > too, but somewhat orthogonal to my point (just standard build on OS > X). But useful to know that's why you don't have G95 showing up, > since you didn't use Sage's provided binary to start with. > > - kcrisman -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org